|Home | About | Journals | Submit | Contact Us | Français|
Human interaction as environmental enrichment for chimpanzees (Pan troglodytes) and other primates is widely promoted and believed to be of value, but has been subject to little objective evaluation. This study assessed the effects of positive human interaction (eg relaxed treat feeding, playing, and other forms of social interaction compatible with personnel safety) on the behaviour of adult chimpanzees. Subjects were housed indoors in groups of two or three individuals. The level of interaction during routine care and management (ie in the process of cleaning, feeding, and monitoring) represented the baseline condition. The test condition involved a familiar caretaker spending an additional 10 minutes per day, 5 days a week, with each chimpanzee. This study was designed to assess carry-over effects of interaction on behaviour outside of the context of care staff presence. Therefore, in all phases of the study, data (97 hours of focal animal sampling) were collected only when caretakers were absent from the building. During the increased human interaction phase, the chimpanzees groomed each other more and showed lower levels of the following behaviours: regurgitation/reingestion, other oral abnormal behaviours, inactivity, and reactivity to the displays of neighboring groups. A trend toward reduced agonistic displaying was detected as well. Attempted interactions with the observer shifted significantly from predominantly aggressive to predominantly affiliative in nature. These results suggest that simple, unstructured affiliation between humans and chimpanzees should be a valued component of behavioural management.
For the purposes of refining the behavioural management of chimpanzees (Pan troglodytes), the effects of varying the physical and conspecific social environment have been investigated in some detail. In comparison to these factors, the amount and manner of human interaction with chimpanzees is a relatively neglected variable in behavioural management. Those humans who work with captive chimpanzees are familiar with the frequent and varied social interchanges between the two species. Particularly when working with chimpanzees in relatively restricted social housing conditions such as single caging and pairs, caregivers intuitively perceive their status as potential social partners for chimpanzees.
The relatively widespread opinion that human interaction is a means to enhance captive environments, and recommendations thereupon (eg Markowitz & Spinelli 1986; Wolfe 1987, 1996; Novak & Drewsen 1989; Bennett 1990; Mahoney 1992; National Research Council 1996), have been subject to relatively little objective evaluation in any taxa. With the exception of anecdotal reports, there have been very few objective evaluations of the effect of human interaction on wellbeing. Bayne et al (1993) found that human interaction associated with treat provisioning resulted in reduced abnormal behaviour in singly-housed rhesus macaques. Positive effects of human interaction as enrichment for great apes have been reported once, by Bloomsmith et al (1999). Both positive reinforcement training and less structured interaction were found to confer benefits to chimpanzees, although the types of behaviours affected varied with the style of interaction provided. Training appeared to benefit social behaviour more broadly, while non-training also ameliorated stereotypic and anxiety-related behaviours (Bloomsmith et al 1999).
Another body of research concerning the effect of human interaction on behaviour relates to positive reinforcement training as a management style. In monkeys, training has been used to foster cooperation during clinical, management, and research procedures (for reviews see Reinhardt 1997a, b; see also Perlman et al 2000). In great apes, training has been used to mitigate social aggression during feeding (Bloomsmith et al 1994) and is effective for fostering cooperation for management and research routines, such as sample collection (Laule et al 1996; Brown & Loskutoff 1998; Lambeth et al 2000; Perlman et al 2001) and transfer between enclosures (Kessel-Davenport & Gutierrez 1994; Bloomsmith et al 1998). Although clearly relevant to reduced stress by obviating the need for invasive and/or aversive techniques such as darting and squeeze-boxing, the effect of training (ie the shift from coercive to cooperative management) on overall welfare remains to be demonstrated.
One other way in which humans influence primate wellbeing is through their passive presence. Primates that are generally considered habituated to people may in fact still be responding negatively. For example, species subject to heavy predation pressure in the wild may persist in performing anti-predator strategies in response to familiar people (Caine 1992). Routine monitoring or observation of macaques by familiar personnel results in persistent stress responses (Malinow et al 1974; Manuck et al 1983; Line et al 1989). Chimpanzees fight and wound each other more frequently during those times when facilities see more human traffic and activities (Maki et al 1987; Lambeth et al 1997). Group dynamics in zoo chimpanzees and other species are generally affected in a negative fashion by the presence of large crowds of visitors, which is associated with lower levels of affiliative behaviour and increased aggression (reviewed in Hosey 2000). Negative effects of noisy zoo visitors have been observed in orang-utans as well (Birke 2002).
The negative effects that the presence of people can have on captive primates underscores the importance of careful evaluation of human interaction as enrichment, no matter how intuitive the apparent benefit. The present study quantifies the effect of increased positive human interaction on the behavioural wellbeing of chimpanzees. This study involved a set of pair- and trio-housed chimpanzees, housed in one building, that had previously been studied in investigations of single vs. small-group housing (Baker 1996) and bedding/forage as enrichment (Baker 1997). The latter study provides the opportunity to contrast the behavioural effects of two different interventions with the same subjects and housing.
Subjects included seven female and five male adult chimpanzees (Pan troglodytes) housed in the Chimpanzee Infectious Disease building at the Yerkes National Primate Research Center in Atlanta, Georgia, USA (See Table 1). All study subjects had been challenged with HIV, but none showed clinical symptoms during the period of data collection. All individuals were nursery-reared except for one mother-reared male. Six individuals were housed in pairs (one male/female, two male/male), and six in single-sex trios. A thirteenth individual, a member of an all-male trio, was not included as a subject in the study. Midway through the test condition, he was separated from his trio due to serious illness. Therefore, one single-sex trio was reduced to a pair. The social setting of all other subjects remained constant throughout the study period.
All individuals were housed in one building containing interconnecting enclosures with chain link ceilings and cage fronts and solid cement flooring and side walls. Each social group was housed in two or three interconnecting enclosures measuring 3.7 × 3.1 × 2.9 m. Available space per individual ranged from 24.4 to 36.6 m3. Individuals were able to view other groups across a central corridor. Enclosures contained resting boards and several portable objects. Twice-daily meals included commercial biscuits accompanied by varied produce or other feeding enrichment at least daily, with water available ad libitum.
The baseline phase involved twice-daily caretaking visits by care staff. Total care staff presence in the building averaged approximately 2.25 hours per day, seven days a week, as it had for at least the previous two years. The test phase also involved twice-daily caretaking visits, but with care staff presence in the building increased to 4.25 hours per day due to interaction between one caretaker and the chimpanzees; this averaged to 10 minutes of extra interaction with each chimpanzee five days per week. The caretaker received no specific instructions associated with participation in the study. Interactions consisted of unstructured bouts of play, grooming, treat feeding and talking, contingent only upon what the chimpanzees initiated and restricted only by the need for personnel safety. The order of interaction between individuals and duration of interaction on a daily basis resulted from the individual's personal judgment and empathy rather than a predetermined schedule.
A total of 97 hours of data was collected for this study, approximately 5 hours per subject during baseline and 3 hours during test conditions. After baseline data were collected, a straw/forage intervention was conducted (for details see Baker 1997). This intervention was concluded six months before the onset of additional human interaction. Data collection during the human interaction phase was begun three months after the onset of the phase. During all phases, data were collected between the twice-daily caretaker visits, at least one half-hour after the caretaker had left the building. No data were collected when the caretaker was present. Therefore the carry-over effects of human interaction, rather than behavioural effects during visits, was assessed.
Data collection methods were identical to those used in the prior evaluation of straw and forage material (Baker 1997). An instantaneous point-sampling technique (Altmann 1974) with five-minute focal animal test sessions and a 15-second intersample interval was supplemented with ad libitum recording of aggressive interactions and other behaviours of short duration. Individuals were observed according to a pre-defined schedule between the hours of 1100 and 1500 h.
Table 1 defines the twelve behavioural categories analyzed. Data for each individual were pooled across focal tests, and statistical analyses were performed using individuals’ percentage of samples for each behavioural category in each of the two experimental conditions. In addition, the preponderance of affiliation in human-directed behaviour was calculated by dividing the number of samples of this behaviour by the number of samples of all human-directed behaviour (both affiliative and aggressive). Subjects’ reactivity was measured by calculating the percent of neighbor vocalizations and displays that were followed within five seconds by the subject vocalizing or displaying. This reaction represents a normal chimpanzee response (Baker & Aureli 1996), but nevertheless is of concern because it is associated with increased intragroup aggression, and is an apparent source of stress to socially-housed captive chimpanzees (Baker & Aureli 1996; 1997).
Two-tailed Wilcoxon matched-pair signed-rank tests were used for comparing scores in the baseline and test conditions, and alpha was set at 0.05. A P-value between 0.05and 0.10 was reported as a trend.
Removal of data collected on the two males whose social group was altered during the study did not alter the direction of change in any behaviour category analysed and are included in the reported results. Baseline results reported in this study vary from those previously published (Baker 1997; Baker & Easley 1997) due to the exclusion of one subject from the present study (see methods).
Figure 1 shows levels of abnormal and anxiety-related behaviour in the baseline and extra human attention phases of the study. Both regurgitation/reingestion (n = 11; T = 9; P < 0.05) and other abnormal behaviours with oral components (n = 10; T = 3; P < 0.05) were significantly reduced during the period of additional human interaction. Non-oral abnormal behaviours, already at very low levels (0.9 ± 0.4% of samples) were not significantly affected (n = 10; T = 28; ns). While levels of scratching (n=12; T = 39; ns) were not affected by the test condition, there was a trend toward significance for agonistic display (n = 11, T = 12; P < 0.06); level fell during the period of extra human interaction.
Subjects spent significantly less time inactive during test conditions (n = 12; T = 7; P < 0.05) and groomed each other more (n = 12; T = 3; P < 0.05); levels of grooming almost doubled (See Figure 2). Aggressive interactions occurred too rarely to permit statistic analysis.
Most of the chimpanzees spent a considerable amount of time attempting to interact with the data collector (5.5 ± 2.8 % of samples), who was strictly unresponsive. While the reduction in these behaviours failed to reach statistical significance for either affiliative (n = 12; T = 29; ns) or aggressive behaviour (n = 9; T = 9; ns), the proportion of social behaviour that was affiliative in nature increased considerably (n = 11; T = 6; P < 0.05; see Figure 3).
During the phase of extra human interaction, study subjects were significantly less reactive to the vocalizations and displays of chimpanzees in other social groups (n = 11; T = 11; P < 0.05; see Figure 3).
A number of positive behavioural changes were observed when chimpanzees housed indoors and in small social groups were provided additional positive human interaction. Levels of abnormal behaviour fell, subjects were less tense and reactive by several measures, spent less time idle, and engaged in higher levels of affiliative behaviour. These findings are a precise opposite of the effects of passive presence on chimpanzees (Maki et al 1987; Lambeth et al 1997; Wood 1998).
However, this study does not address the behaviour of the chimpanzees when the interactor was present. Bloomsmith et al (1999) found a decrease in grooming and an increases of aggression in the presence of the interactor, likely attributable to competition for attention. Increases in abnormal and anxiety-related behaviours were found as well. The present study involved no data collection during interaction. However, even if subjects in the present study showed undesirable changes in behaviour during the 10 minutes per day of human interaction, this effect is of minimal significance given the benefits to their behaviour during the bulk of their day. It would nonetheless have been helpful to have information on behaviour during interactions, given the contrast in behaviour during versus outside the interaction period in the study by Bloomsmith et al (1999).
One important application of this information would be for informal assessments during implementation. Many forms of enrichment show behavioural benefits during and immediately following provisioning, but, if measured at all, behavioural responses may not persist after the enrichment is no longer present (eg Bryant et al 1988; Bayne et al 1992). It is reasonable to expect that people may use the reaction of the chimps during human interaction as a measure of the overall effect of the intervention. What the interactor perceives as her effect on the chimpanzees’ behaviour may be unrelated to, or even the opposite of, the actual effect outside visits. In other words, one should not judge the effectiveness of this technique solely by what occurs during interactions.
Whereas previously only positive-reinforcement training has been demonstrated to be effective for improving social dynamics (Bloomsmith et al 1994; 1999), in the present study unstructured human interaction benefited social behaviour in several ways. The chimpanzees not only groomed each other at higher levels but engaged in non-contact aggressive interactions less and showed less aggressive ‘contagion’ when others displayed. Anecdotal observations of potentially agitating events (eg visits by unfamiliar people to the building, prolonged activity in the anteroom without entry to the animal area [personal observation]) are in line with this objective measure of decreased reactivity. This shift may have promoted more relaxed relationships within social groups. This is clearly a positive change, since levels of grooming are generally higher in more physically and socially enriched groups of captive chimpanzees. That unstructured interaction in addition to training can benefit social dynamics in chimpanzees is important to note since a wider pool of employees can be drawn from to supply a form of interaction requiring less expertise. As a matter of fact, the caregiver involved in the present study received no specific instructions or schedule for interacting with the animals, and took the lead of the chimpanzees themselves to determine the sequence and style of interaction. The absence of rigorous rules or instruction for interacting with the chimpanzees make the results of this study relevant to a situation in which carestaff are provided time and support for interacting with the chimpanzees. It should be noted, however, that for personnel safety, as well as the benefit of the chimpanzees, individuals interacting with chimpanzees should be familiar not only to the species but to the individual chimpanzees involved.
The present study found positive effects on more classes of behaviours than did Bloomsmith et al (1999). In Bloomsmith et al (1999) reductions in abnormal behaviour and aggression were observed. In the present study, in addition to these same results, reduced inactivity and increased grooming were seen as well. There are several possible explanations for the contrast in findings. First, the current study involved chimpanzees housed only in pairs and trios, while Bloomsmith et al (1999) included study subjects housed in groups numbering from two to seven. It is possible that the higher level of conspecific social opportunities for many subjects decreased the impact that additional social interaction with humans would have. Second, while the subjects in Bloomsmith et al (1999) were housed in indoor-outdoor runs, the subjects in the present study had no access to the outdoors and showed lower baseline levels of wellbeing than a comparable cohort housed with outdoor access (Baker & Ross 1998). If the baseline level of environmental complexity is lower, the response to an environmental enhancement may be more pronounced. Third, the subjects in the present study were predominantly nursery-reared, while the background of subjects in Bloomsmith et al (1999) was more varied. It is likely that rearing by humans, even when housed with conspecific peers, influences reaction to people in adulthood. Fourth, the social setting remained consistent for all subjects used in Bloomsmith et al (1999) while two subjects in the present study experienced a reduction in group size during the study. This represents a potential confound, but is unlikely to be responsible for the types of behaviours altered and direction of change in behaviours. Last, the added amount of human interaction provided to the chimpanzees was smaller in the study by Bloomsmith et al (1999), involving sixty minutes per week of extra interaction to each social group, regardless of its size. In the present study, groups received 60-150 minutes of interaction, depending on its size. Because so many variables differed between studies, their comparison does not suggest that the larger amounts of interaction are of no added benefit. Controlled comparisons of different quantities of human interaction are badly needed in order to determine realistically-implemented manpower efforts with maximal benefit to chimpanzee welfare. It would also be valuable to compare the effect of increased interaction implemented by one person with increased interaction through visits by a number of individuals, since the latter represents an especially practical means for boosting overall interaction time with humans.
The results of this study are surprisingly similar to the effects of straw and forage material, a different class of enrichment, on this same population. Providing constant opportunities for foraging was hypothesized to be the most promising intervention to reduce the high levels of appetitive abnormal behaviours in this population (Baker & Easley 1996; Baker 1997). However, while these abnormal behaviours seemed clearly tied to increasing hunger, social enrichment appeared equally effective in ameliorating these behaviours; in fact, levels of regurgitation/reingestion and other oral abnormal behaviours were indistinguishable between interventions. Also similar were the decreases in agonistic displaying and increased activity. Affiliative behaviour also increased in the presence of straw and forage. However, in that study it was play rather than grooming that showed an increase, probably related to the opportunities for play in the presence of straw, which was frequently incorporated in play sessions (Baker 1997). Methodologically, the comparison between the two studies calls into question the necessity of tailoring interventions to the precise nature of the behavioural deficits found in an individual. While tailoring enrichment to the underlying motivation for undesirable behaviours can often be productive (eg Carlstead & Seidensticker 1991), one may not in fact need to restrict oneself to interventions strictly related to the class of behaviour that is problematic.
Reaction to the observer is perhaps an underused measure in behavioural management studies. A few exceptions include a comparison of chimpanzee housing configurations (Rice et al 1999) and the use of alarm vocalizations in evaluating levels of stress in capuchins (Boinski et al 1999). The ability to compare responses to people with other measures of wellbeing is important for validating the use of this behaviour as a measure of wellbeing. For example, Boinski et al (1999) found concomitant reductions in cortisol and abnormal behaviour in response to increased inanimate enrichment. Singly-housed chimpanzees are aggressive to observers more frequently and show higher levels of aggressive and anxiety-related behaviours than socially housed chimpanzees (Baker 1996). In the present study, the switch from predominantly aggressive to predominantly affiliative attempts to engage the observer was accompanied by several other positive behavioural changes. This finding supports the use of reaction to people as a measure in chimpanzee management studies.
Adding 50 minutes per week of positive human attention to the management of chimpanzees housed indoors in pairs and trios resulted in improved welfare by several measures: reduced abnormal behaviour, tension-related behaviour, inactivity, aggressivity to the observer, and reactivity to the agonistic displays of others, and increased social grooming between conspecifics, all during periods when the interactor was not present. These results confirm the often perceived notion that human interaction is valued by captive chimpanzees, and the often recommended notion that human interaction should be valued as a part of behavioural management.
This study could not have been accomplished without the efforts and dedication of the Yerkes Animal Care Staff. This research was supported by NIH Grant RR-00165 to the Yerkes Regional Primate Research Center. Yerkes is fully accredited by the American Association for Accreditation of Laboratory Animal Care.